Mohan Kankanhalli

VidHal: Benchmarking Hallucinations in Vision LLMs

Translution: Unifying Self-attention and Convolution for Adaptive and Relative Modeling

Involuntary Jailbreak

Fair Deepfake Detectors Can Generalize

Strong Preferences Affect the Robustness of Preference Models and Value Alignment

The VLLM Safety Paradox: Dual Ease in Jailbreak Attack and Defense

UnSTAR: Unlearning with Self-Taught Anti-Sample Reasoning for LLMs

SCAN: Bootstrapping Contrastive Pre-training for Data Efficiency

Towards Robust Concept Erasure in Diffusion Models: Unlearning Identity, Nudity and Artistic Styles

A Semantic Data Augmentation driven Nested Adapter for Video Moment Retrieval